Breakout Discussion Topics List


Discussion Leader - Topic

  1. Andrew Leaver-Fay - Rosetta Energy Function � Scientific Benchmarking, What is Still Missing? (Outside Chapel)
  2. Sarel Fleishman - Modeling/Designing Antibodies (Kingfisher Meadow)
    Antibodies are equally exciting and frustrating to model. Exciting because of the very broad range of applications and frustrating because they use loops to mediate function bringing up tradeoffs between stability, rigidity, and function that protein design/modeling has traditionally found difficult to resolve. Addressing these challenges is key to modeling antibodies and we'll discuss these questions and possible routes to solving them. The general direction our lab has taken on antibody modeling has been to exploit the structural modularity in antibodies. If you want background on the antibody architecture and sequence/structure relationships ahead of the meeting perhaps the most fundamental paper is Chothia's JMB 273:927.

    As a tentative plan we'll devote half of the discussion to design and half to modeling. If people are interested in other antibody related questions (thermostabilization, specificity switching, species specificity, humanization/murinization, metamorphism), please email me at: [email protected] and we'll accommodate these questions during the meeting.

  3. Kevin Drew - Peptidomimetics and non-canonical amino acids (Kingfisher Meadow)
    • July 31st General - possible topics: Cyclic peptides/peptoids/RNA/saccharides (compatible types of cyclization: click, crosslinks, thioalkylation), New non-canonical amino acids (AAs) or backbones (BBs) to include (N-me AAs, alpha-hydroxy acids, etc?), best practices (energy functions, flags, adding new AAs/BBs etc), design mixed backbones (on the fly?)
    • August 1st Technical - possible topics: residue type bloat, torsion specification in params file (action?), other peptide-centric code that needs to be generalized? coverage of integration/unit tests, are all parallel efforts compatiable? (do we know what everyone else is doing so we are not duplicating code?)
  4. Neil King - Protein Assemblies and protein-based materials (Green Room)
  5. Jane Richardson & John Karanicolas - RNA and RNA-protein interactions (Grotto Pub) - second session (Aug 1) only!!

    The direction of the discussion will be dictated by the interests of the attendees. These may include: comparative modeling of RNAs and protein-RNA complexes, RNA backbone conformers, moves for sampling RNA conformational space, energy function considerations, motif-based design, and small-molecule inhibitors.

  6. Karen Fleming - Membrane Proteins (Kingfisher Meadow)
  7. Dominick Gront - Comparative Modeling (Quail Deck)
  8. Justin Siegel Enzyme design and small molecule docking (Salmon Gallery Terrace)
  9. Eva-Maria Strauch Design of protein-protein interactions (Salmon Gallery)
  10. Nik Sgourakis - Structural Biology � integrating experimental data with Rosetta modeling (Flicker)
    • 1st session: Examples by discussion members (eg. cryo-EM, NMR, H/D exchange, fiber diffraction, scattering and cross-linking techniques). Applications on modeling protein complexes and protein-nucleic acid interactions.
    • 2nd session: Standards for cross-validating our models. What can we learn from interacting with different datasets (not just X-ray) towards improving the Rosetta energy function?
  11. Ryan Hallet - High throughput screening integration with Rosetta (Nuthatch)
  12. Ora Furman - Modeling and Design of Specificity (Dipper)
  13. Firas Khatib � Designing Rosetta-based science puzzles: Is there wisdom in the crowds? (TBD)
  14. Colin Smith � Predicting and designing protein motions (TBD)

    Possible questions: What triggers the motions one would want to predict or design? (For example: Nothing, they just happen free in solution. Light. Covalent modification. Binding event.) How can one predict motions? (For example: Full sampling of possible states. Heuristics based on sampling in and around different minima.) What level of prediction accuracy is necessary for it to be useful? (For example: Just in the rough ballpark. It a fools errand, prediction will never be useful.) What heuristics for design would be useful? (For example: Only model two states, the others don't matter. There's no cutting corners, you have to do a full prediction for each sequence.)

HACK